class WorkflowExportSerializer(serializers.ModelSerializer):
"""
This serializer is use to export Workflows selecting a subset of
actions. Since the SerializerMethodField used for the selection is a
read_only field, the import is managed by a different serializer that
uses a regular one for the action field (see WorkflowImportSerializer)
"""
actions = serializers.SerializerMethodField('get_filtered_actions')
data_frame = DataFramePandasField(
required=False,
help_text=_('This field must be the Base64 encoded '
'result of pandas.to_pickle() function'))
columns = ColumnSerializer(many=True, required=False)
views = ViewSerializer(many=True, required=False)
version = serializers.CharField(read_only=True,
default='NO VERSION',
allow_blank=True,
label="OnTask Version",
help_text=_("To guarantee compability"))
def get_filtered_actions(self, workflow):
# Get the subset of actions specified in the context
action_list = self.context.get('selected_actions', [])
if not action_list:
# No action needs to be included, no need to call the action
# serializer
return []
# Execute the query set
query_set = workflow.actions.filter(id__in=action_list)
# Serialize the content and return data
serializer = ActionSerializer(instance=query_set,
many=True,
required=False)
return serializer.data
def create(self, validated_data, **kwargs):
# Initial values
workflow_obj = None
try:
workflow_obj = Workflow(
user=self.context['user'],
name=self.context['name'],
description_text=validated_data['description_text'],
nrows=0,
ncols=0,
attributes=validated_data['attributes'],
query_builder_ops=validated_data.get('query_builder_ops', {}))
workflow_obj.save()
# Create the columns
column_data = ColumnSerializer(data=validated_data.get(
'columns', []),
many=True,
context={'workflow': workflow_obj})
# And save its content
if column_data.is_valid():
column_data.save()
else:
raise Exception(_('Unable to save column information'))
# If there is any column with position = 0, recompute (this is to
# guarantee backward compatibility.
if workflow_obj.columns.filter(position=0).exists():
for idx, c in enumerate(workflow_obj.columns.all()):
c.position = idx + 1
c.save()
# Load the data frame
data_frame = validated_data.get('data_frame', None)
if data_frame is not None:
ops.store_dataframe_in_db(data_frame,
workflow_obj.id,
reset_keys=False)
# Reconcile now the information in workflow and columns with the
# one loaded
workflow_obj.data_frame_table_name = \
pandas_db.create_table_name(workflow_obj.pk)
workflow_obj.ncols = validated_data['ncols']
workflow_obj.nrows = validated_data['nrows']
workflow_obj.save()
# Create the actions pointing to the workflow
action_data = ActionSerializer(data=validated_data.get(
'actions', []),
many=True,
context={'workflow': workflow_obj})
if action_data.is_valid():
action_data.save()
else:
raise Exception(_('Unable to save column information'))
# Create the views pointing to the workflow
view_data = ViewSerializer(data=validated_data.get('views', []),
many=True,
context={'workflow': workflow_obj})
if view_data.is_valid():
view_data.save()
else:
raise Exception(_('Unable to save column information'))
except Exception:
# Get rid of the objects created
if workflow_obj:
if workflow_obj.has_data_frame():
pandas_db.delete_table(workflow_obj.id)
if workflow_obj.id:
workflow_obj.delete()
raise
return workflow_obj
class Meta(object):
model = Workflow
# fields = ('description_text', 'nrows', 'ncols', 'attributes',
# 'query_builder_ops', 'columns', 'data_frame', 'actions')
exclude = ('id', 'user', 'created', 'modified',
'data_frame_table_name', 'session_key', 'shared')
def create(self, validated_data, **kwargs):
# Initial values
workflow_obj = None
try:
workflow_obj = Workflow(
user=self.context['user'],
name=self.context['name'],
description_text=validated_data['description_text'],
nrows=0,
ncols=0,
attributes=validated_data['attributes'],
query_builder_ops=validated_data.get('query_builder_ops', {}))
workflow_obj.save()
# Create the columns
column_data = ColumnSerializer(data=validated_data.get(
'columns', []),
many=True,
context={'workflow': workflow_obj})
# And save its content
if column_data.is_valid():
column_data.save()
else:
raise Exception(_('Unable to save column information'))
# If there is any column with position = 0, recompute (this is to
# guarantee backward compatibility.
if workflow_obj.columns.filter(position=0).exists():
for idx, c in enumerate(workflow_obj.columns.all()):
c.position = idx + 1
c.save()
# Load the data frame
data_frame = validated_data.get('data_frame', None)
if data_frame is not None:
ops.store_dataframe_in_db(data_frame,
workflow_obj.id,
reset_keys=False)
# Reconcile now the information in workflow and columns with the
# one loaded
workflow_obj.data_frame_table_name = \
pandas_db.create_table_name(workflow_obj.pk)
workflow_obj.ncols = validated_data['ncols']
workflow_obj.nrows = validated_data['nrows']
workflow_obj.save()
# Create the actions pointing to the workflow
action_data = ActionSerializer(data=validated_data.get(
'actions', []),
many=True,
context={'workflow': workflow_obj})
if action_data.is_valid():
action_data.save()
else:
raise Exception(_('Unable to save column information'))
# Create the views pointing to the workflow
view_data = ViewSerializer(data=validated_data.get('views', []),
many=True,
context={'workflow': workflow_obj})
if view_data.is_valid():
view_data.save()
else:
raise Exception(_('Unable to save column information'))
except Exception:
# Get rid of the objects created
if workflow_obj:
if workflow_obj.has_data_frame():
pandas_db.delete_table(workflow_obj.id)
if workflow_obj.id:
workflow_obj.delete()
raise
return workflow_obj
def create(self, validated_data, **kwargs):
# Process first the used_columns field to get a sense of how many
# columns, their type how many of them new, etc. etc.
new_columns = []
for citem in validated_data['used_columns']:
cname = citem.get('name', None)
if not cname:
raise Exception(_('Incorrect column name {0}.').format(cname))
col = Column.objects.filter(workflow=self.context['workflow'],
name=cname).first()
if not col:
# new column
if citem['is_key']:
raise Exception(
_('New action cannot have non-existing key '
'column {0}').format(cname))
# Accummulate the new columns just in case we have to undo
# the changes
new_columns.append(citem)
continue
# existing column
if col.data_type != citem.get('data_type', None) or \
col.is_key != citem['is_key'] or \
set(col.categories) != set(citem['categories']):
# The two columns are different
raise Exception(
_('Imported column {0} is different from existing '
'one.').format(cname))
# Update the column categories (just in case the new one has a
# different order)
col.set_categories(citem['categories'])
new_column_names = [x['name'] for x in new_columns]
action_type = validated_data.get('action_type', None)
if not action_type:
if validated_data.get('is_out'):
action_type = Action.PERSONALIZED_TEXT
else:
action_type = Action.SURVEY
action_obj = None
try:
# used_columns has been verified.
action_obj = Action(
workflow=self.context['workflow'],
name=validated_data['name'],
description_text=validated_data['description_text'],
action_type=action_type,
serve_enabled=validated_data['serve_enabled'],
active_from=validated_data['active_from'],
active_to=validated_data['active_to'],
content=validated_data.get('content', ''),
target_url=validated_data.get('target_url', None),
shuffle=validated_data.get('shuffle', False))
action_obj.save()
if new_columns:
# There are some new columns that need to be created
column_data = ColumnSerializer(data=new_columns,
many=True,
context=self.context)
# And save its content
if column_data.is_valid():
column_data.save()
workflow = self.context['workflow']
df = pandas_db.load_from_db(self.context['workflow'].id)
if df is None:
# If there is no data frame, there is no point on
# adding columns.
Column.objects.filter(
workflow=self.context['workflow'],
name__in=new_column_names).delete()
action_obj.delete()
raise Exception(
_('Action cannot be imported with and '
'empty data table'))
for col in Column.objects.filter(
workflow=workflow, name__in=new_column_names):
# Add the column with the initial value
df = ops.data_frame_add_column(df, col, None)
# Update the column position
col.position = len(df.columns)
col.save()
# Store the df to DB
ops.store_dataframe_in_db(df, workflow.id)
else:
raise Exception(_('Unable to create column data'))
# Load the conditions pointing to the action
condition_data = ConditionSerializer(
data=validated_data.get('conditions', []),
many=True,
context={'action': action_obj})
if condition_data.is_valid():
condition_data.save()
else:
raise Exception(_('Unable to create condition information'))
# Update the condition variables for each formula if not present
for condition in action_obj.conditions.all():
if condition.columns.all().count() == 0:
col_names = get_variables(condition.formula)
# Add the corresponding columns to the condition
condition.columns.set(
self.context['workflow'].columns.filter(
name__in=col_names))
# Load the columns field
columns = ColumnNameSerializer(data=validated_data['columns'],
many=True,
required=False,
context=self.context)
if columns.is_valid():
for citem in columns.data:
column = action_obj.workflow.columns.get(
name=citem['name'])
action_obj.columns.add(column)
columns.save()
else:
raise Exception(_('Unable to create columns field'))
except Exception:
if action_obj and action_obj.id:
action_obj.delete()
Column.objects.filter(workflow=self.context['workflow'],
name__in=new_column_names).delete()
raise
return action_obj